DEveloping Tests for Endometrial Cancer deTection (DETECT)

Abstract

Background: Postmenopausal bleeding (PMB) is the 'red flag' symptom for endometrial cancer and prompts a sequence of investigations for women that are invasive, costly, and often painful and distressing. Most women with PMB do not have cancer and undergo these investigations unnecessarily. A simple triage tool to identify those at highest risk of having endometrial cancer and reassure those at low risk would transform the diagnostic pathway. Methods: A prospective, multicentre study (DETECT) was performed in which voided urine and Delphi-screener collected vaginal samples were obtained from consecutive women undergoing clinical investigations for PMB. Samples were assessed by two independent cytologists blinded to cancer status, and the results compared to histopathology results to determine diagnostic test accuracy. Patient demographics, metabolic and reproductive risk factors, urinalysis, and transvaginal scan (TVS) results were compared between endometrial cancer cases and controls in uni- and multivariable analyses. These data were used to validate published endometrial cancer diagnostic models and to create novel models to triage women for further investigations, either before (Community model) or after TVS (Scan model). Results: Of 1864 participants, 115 (6.17%) had endometrial (n=99) or other pelvic malignancies (n=16). The sensitivity and specificity of combined urine and vaginal cytology were 80.8% (95% CI: 71.7-88.0%) and 92.6% (95% CI: 91.2-93.8%) for endometrial cancer detection and 80.0% (95% CI: 71.5-86.9%) and 92.6% (95% CI: 91.2-93.8%) for any pelvic cancer detection. Risk factors associated with endometrial cancer included age [OR 1.11 per one year increase in age (95% CI 1.09-1.14)], body mass index (BMI) [OR 1.03 per 1kg/m2 increase in BMI (95% CI 1.01-1.06)], recurrent vaginal bleeding [OR 6.00 (95% CI 3.26-11.05)], and endometrial thickness on TVS [OR of 1.22 (95% CI 1.18-1.26) per 1mm increase]. Haematuria, leukocyturia and proteinuria were all strongly associated with endometrial cancer, with OR of 5.42 (95% CI 3.53-8.32), 3.42 (95% CI 2.07-5.17) and 6.17 (95% CI 2.70-14.07) respectively. The best performing published scan model gave an area under the receiver operating characteristic curve (AUC) of 0.92 (95% CI 0.90-0.95). Our novel Community model gave an AUC of 0.879 (95% CI 0.847 to 0.910) and our Scan model an AUC of 0.940 (95% CI 0.921 to 0.958), with both showing clinical utility. Conclusion: Urogenital cytology is a minimally invasive test that holds great promise for the detection of endometrial cancer, however, studies exploring its use in clinical practice are needed. Novel diagnostic models incorporating haematuria performed better than previous comparable models and showed clinical utility, although external validation in a larger cohort is needed.

Date of Award	1 Aug 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Emma Crosbie (Supervisor), Jamie Sergeant (Supervisor) & Richard Edmondson (Supervisor)